An academic survey involving researchers from multiple universities and industry institutions suggests that, despite the recent surge in interest in "AI-themed tokens" and decentralized AI projects, the substantive integration of cryptocurrency and AI remains in its early stages. The paper argues that many current projects are still driven primarily by narratives, with genuine economic advantages and mainstream adoption yet to be fully validated.
The research divides the track into two categories.
Researchers note that the market often groups all "crypto + AI" projects into a single category, but the technological approaches and business assumptions of these two types of projects differ. The paper divides this intersection into two parts: "Crypto x AI" and "AI x Crypto".
The former involves using AI to enhance cryptographic systems, including fraud detection, smart contract analysis, on-chain data analysis, and AI-assisted protocol development. The latter involves using blockchain infrastructure to support AI systems, encompassing decentralized AI infrastructure, verifiable AI, privacy-preserving computation, and AI agent payments.
Decentralized AI faces a cost test
The primary criticism in the paper focuses on the economic viability of decentralized AI infrastructure. The authors argue that while many projects have demonstrated technical feasibility, they have not yet sufficiently proven superiority over centralized service providers in terms of cost, efficiency, and commercial viability.
This means that projects centered around decentralized GPU markets, distributed computing networks, and AI-oriented DePIN may face stricter real-world scrutiny going forward. The study notes that openness and censorship resistance are advantages of decentralized systems, but these characteristics do not automatically translate into competitive commercial products.
AI agents are highlighted as a promising payment solution.
Although the overall tone is cautious, the paper does not dismiss the entire potential of combining cryptography and AI. Researchers believe that blockchain payment networks and stablecoins could provide clear practical applications for AI systems.
Among the key areas highlighted is autonomous trading between AI agents. When AI agents need to invoke services, purchase data, or pay for computational power without human intervention, on-chain payment tools may be more effective.
New risks emerge simultaneously
The study also warns that integrating AI with crypto infrastructure introduces new attack surfaces and governance challenges. Risks highlighted in the paper include runaway autonomous agents, AI-controlled malicious smart contracts, and privacy conflicts within decentralized AI systems.
Overall, this research does not deny the long-term potential of combining crypto and AI, but suggests that current market enthusiasm has outpaced real-world applications. For related projects, the next critical step is to demonstrate cost advantages, user demand, and practical use cases.